Mutual Diffusion Coefficients and Densities at 298.15 K of Aqueous

mixtures of NaCl and Na2SO4 at a constant total molarity of 1.500 mol‚dm-3 at 298.15 K, using .... The values z1 ) 1 and z1 ) 0 denote the limiting ...
0 downloads 0 Views 129KB Size
936

J. Chem. Eng. Data 2000, 45, 936-945

Mutual Diffusion Coefficients and Densities at 298.15 K of Aqueous Mixtures of NaCl and Na2SO4 for Six Different Solute Fractions at a Total Molarity of 1.500 mol‚dm-3 and of Aqueous Na2SO4 Onofrio Annunziata,† Joseh A. Rard,*,| John G. Albright,‡ Luigi Paduano,§ and Donald G. Miller⊥ Chemistry Department, Texas Christian University, Fort Worth, Texas 76129, and Geosciences and Environmental Technologies, Earth and Environmental Sciences Directorate, Lawrence Livermore National Laboratory, University of California, Livermore, California 94550

Isothermal mutual diffusion coefficients (interdiffusion coefficients) were measured for ternary aqueous mixtures of NaCl and Na2SO4 at a constant total molarity of 1.500 mol‚dm-3 at 298.15 K, using Rayleigh interferometry with computerized data acquisition. The experiments were performed at NaCl molarity fractions of z1 ) 1, 0.90, 0.75, 0.50, 0.25, and 0. These measurements supplement our earlier results at total molarities of 0.500 and 1.000 mol‚dm-3. Diffusion coefficients were also measured at three additional concentrations of Na2SO4(aq). Densities of all solutions were measured with a vibrating tube densimeter. At all ternary solution compositions, one cross-term diffusion coefficient has small negative values whereas the other has larger positive values. Comparing the results at 0.500, 1.000, and 1.500 mol‚dm-3 shows that both main-term diffusion coefficients decrease with increasing total concentration at any fixed value of z1, whereas both cross-term coefficients are shifted in a positive direction. At 1.500 mol‚dm-3 both the NaCl main-term and cross-term diffusion coefficients have a maximum as a function of z1, whereas the Na2SO4 cross-term coefficient has a minimum. Trace diffusion coefficients D*(Cl-) ) (1.145 ( 0.02) × 10-9 m2‚s-1 and D*(SO42-) ) (0.805 ( 0.015) × 10-9 m2‚s-1 were extrapolated from these results for the Cl-(aq) ion in 1.500 mol‚dm-3 Na2SO4(aq) and for the SO42-(aq) ion in 1.500 mol‚dm-3 NaCl(aq). Values of D*(Cl-) in Na2SO4(aq) and in NaCl(aq) were found to be essentially identical, as were D*(SO42-) in these same two electrolytes, provided the comparisons are made at the same volumetric ionic strengths.

Introduction Diffusive transport occurs in many chemical, geochemical, and industrial processes,1-4 and diffusion coefficients are needed for calculation of various types of generalized transport coefficients.5-9 Aqueous salt solutions are often used to solubilize proteins, and increasing the salt concentration will sometimes cause salting-out of protein crystals. Various salts, including NaCl and Na2SO4, can produce this salting-out effect. Diffusion data for these salts and their mixtures will complement ongoing work at Texas Christian University to determine diffusion coefficients of ternary and quaternary aqueous solutions containing lysozyme. That work will provide fundamental data for modeling liquid-phase transport during protein crystal growth. There is also a long-standing program at Lawrence Livermore National Laboratory of studying the thermodynamic and transport properties of aqueous solutions of the various salts present in seawater and natural brines and their mixtures. The present report is a continuation of our collaboration to characterize the diffusion coefficients of systems of mutual interest. Felmy and Weare2 have described a method of estimating diffusion coefficients based on Miller’s estimation methods6 and using Onsager’s phenomenological coef†

Texas Christian University. E-mail: [email protected]. Lawrence Livermore National Laboratory. E-mail: [email protected]. Texas Christian University. E-mail: [email protected]. § Permanent address: Dipartimento di Chimica, Universita ` di Napoli, 80134 Napoli, Italy. E-mail: [email protected]. ⊥ Lawrence Livermore National Laboratory. E-mail: [email protected]. |



ficients, with the chemical potential derivatives being calculated using Pitzer’s ion interaction model.10 Felmy and Weare examined available diffusion data for various subsystems derived from the six-ion Na-K-Ca-Mg-ClSO4-H2O seawater model. Diffusion data were lacking for the majority of the ternary solution subsystems, with most of the available diffusion data being for common-ion chloride salt mixtures. Mutual diffusion coefficients have been reported from dilute solution to near saturation at 298.15 K for binary aqueous solutions of many of the major and minor salts present in seawater and other natural waters, and those experimental studies are summarized elsewhere.11 Mutual diffusion data are also available at 298.15 K for several common-anion ternary aqueous solution compositions relevant to natural waters. These are mostly chloride salt mixtures12-25 but include one composition of the system NaHCO3 + KHCO3 + H2O.26 Hao and Leaist27 studied the noncommon-ion mixtures of NaCl + MgSO4 + H2O. Relatively few diffusion studies have been reported for common-cation aqueous mixtures. Diffusion data have been reported for mixtures of NaOH + Na2SO3 + H2O and mixtures of MgCl2 + MgSO4 + H2O by Leaist and coworkers,28,29 and for eight compositions of NaCl + Na2SO4 + H2O at total molarities of 0.500 and 1.000 mol‚dm-3.11,18,30 Except at low molarities of solute, the binary or ternary solution diffusion coefficients D and Dij cannot be predicted quantitatively by the Nernst-Hartley equations,5,6,11,16,17,24 which are based on an infinite dilution model, even when activity coefficient derivative corrections and other factors are included in the calculations.17 Consequently, we are

10.1021/je000134s CCC: $19.00 © 2000 American Chemical Society Published on Web 08/19/2000

Journal of Chemical and Engineering Data, Vol. 45, No. 5, 2000 937 determining accurate Dij values for a few representative ternary and quaternary aqueous salt systems to characterize experimentally the dependences of these coefficients on total concentration and on the solute mole ratio. These experimental Dij can then be used as a “test bed” for developing and refining methods to estimate multicomponent solution Dij for arbitrary mixtures at higher concentrations. Since experimental Dij for common-cation mixtures are quite limited, including those with sodium salts, in 1995 we began a systematic investigation of the Dij for the system NaCl + Na2SO4 + H2O at 298.15 K but interrupted that study because of an interruption in funding. However, we now have support that allows our previous results11,30 at 0.500 and 1.000 mol‚dm-3 to be extended to 1.500 mol‚dm-3. Because of solubility limitations resulting from the precipitation of Na2SO4‚10H2O(cr), 1.500 mol‚dm-3 is close to the maximum concentration for which diffusion coefficient measurements can be made over the full range of z1. The present study was undertaken in part to determine whether the cross-term diffusion coefficients D12 and D21 remain positive and negative, respectively, as the total concentration of the solution is increased to 1.500 mol‚dm-3. Although this was found to be true at 1.500 mol‚dm-3, there were major changes in several of the diffusion coefficients, particularly for D11. Since this report continues our earlier work, and because most details of the experimental measurements and data processing are identical to those reported in the earlier studies, we refer the readers to a previous paper11 for more of these details. Experimental Section All experiments were performed at Texas Christian University. Diffusion Coefficient Measurements. Diffusion experiments were performed by Rayleigh interferometry at (298.15 ( 0.005) K with free-diffusion boundary conditions, using the high-quality Gosting diffusiometer31 with automated data recording. Miller et al.24 and Miller and Albright32 have described the Rayleigh method in considerable detail, including experimental and computational procedures. The molarity fractions z1 of NaCl in these {z1NaCl + (1 - z1)Na2SO4}(aq) mixtures are 1, 0.900 00, 0.749 93, 0.500 00, 0.250 00, and 0 at a total molarity of 1.500 mol‚dm-3. These z1 values are very nearly the same as those used at 0.500 and 1.000 mol‚dm-3,11,30 which facilitates comparisons with our previous diffusion data. The values z1 ) 1 and z1 ) 0 denote the limiting binary solutions NaCl(aq) and Na2SO4(aq), respectively. For each ternary solution composition, four diffusion experiments were performed at essentially the same average concentrations of each solute, C h 1 and C h 2. However, these four experiments involved different values of the ratio ∆C1/(∆C1 + ∆C2), where the ∆Ci are the differences between the concentrations of electrolyte i between the bottom and top sides of the initial diffusion boundary. The subscript 1 denotes NaCl, and the subscript 2 denotes Na2SO4. These ratios were selected to correspond to the refractive index fractions Ri ≈ 0, 0.2, 0.8, and 1, as recommended by Dunlop12 and O’Donnell and Gosting.14 The Ri are given by

Ri ) Ri∆Ci/(R1∆C1 + R2∆C2) ) Ri∆Ci/J

(1)

where J is the total number of Rayleigh interference fringes (which is generally not an integer) and Ri is the refractive

index increment of J with respect to the concentration of solute i. The Ri and Ri are obtained by the method of leastsquares from each set of four experimental J and ∆Ci values measured for the same overall composition in the same diffusion cell. A more fundamental refractive index increment Ri* describes the difference in refractive index ∆n between the top and bottom solutions forming the diffusion boundary by the equation ∆n ) R1*∆C1 + R2*∆C2. These ∆n are directly related to J by ∆n ) λJ/a, where λ ) 543.3655 nm is the wavelength in air of the helium-neon laser green line used by our interferometer and a is the path length of the light inside the diffusion cell. Consequently, R1 ) aR1*/λ and R2 ) aR2*/λ. We report the Ri because J is the directly observed experimental quantity. (The Ri* are not known as precisely because they contain an additional uncertainty arising from the determination of a.) Extraction of the Dij from the fringe position data only requires the ratio Ri/Rj ) Ri*/Rj*, so it is immaterial which type of refractive index increment is reported. All diffusion experiments were performed in cell C-1235H-11, for which a ) 2.5057 cm and which has a magnification factor of 1.76070 at the focal plane of our diffusiometer. This magnification factor was obtained by placing a transparent, precisely ruled scale in the center-of-cell position in the thermostated water bath and then comparing the observed scale line separations at the photodiode array11 to the corresponding separations of those same lines on the original ruled scale as measured with a microcomparator. The reported value of the magnification factor is the average of several separate determinations and is precise to better than 0.03%. A computer-controlled photodiode array was used for the “real time” recording of positions of the Rayleigh fringe patterns during the diffusion experiments. This photodiode array and its operation, the cell-filling techniques, the recording of the baseline patterns and the Rayleigh fringe patterns, and so forth are essentially the same as those described by Rard et al.11 However, the original 66 MHz 486 DX computer was replaced with a 166 MHz Pentium computer to increase the speed of acquiring and processing the experimental information. One computer code is used for interpolation of the experimental fringe position information recorded by the photodiode array to reconstruct the positions of 100 symmetrically paired fringes for each temporal scan, of which 96 of the fringe pairs are used in subsequent calculations. This code generates a file for each individual diffusion experiment that contains the J values, the 96 interpolated symmetrically paired Rayleigh fringe positions for each of the 50 scans for an individual experiment, and the times at which the Rayleigh patterns were recorded. For each set of four experiments at a given overall composition (fixed z1), a second computer program TFIT combines the corresponding files and the concentration differences ∆Ci of each solute across the initial diffusion boundary. This computer program calculates the diffusion coefficients, their standard errors, and other pertinent quantities as described previously.11,24,33 Solution Preparation and Density Measurements. Solutions were prepared by mass from samples of NaCl(cr) that had previously been dried in air at 723 K,34 from samples of stock solutions of Na2SO4(aq), and from purified water. The water purification was described previously.11 The assumed molar masses are 58.443 g‚mol-1 for NaCl, 142.037 g‚mol-1 for Na2SO4, and 18.0153 g‚mol-1 for H2O.

938

Journal of Chemical and Engineering Data, Vol. 45, No. 5, 2000

Table 1. Results from Binary Solution Mutual Diffusion Coefficient and Density Measurements for NaCl(aq) and Na2SO4(aq) Solutions at 298.15 K with Rayleigh Interferometry and Vibrating Tube Densimetrya quantity

NaCl(aq)

Na2SO4(aq) b

Na2SO4(aq) c

Na2SO4(aq) c

Na2SO4(aq) c

Na2SO4(aq)c

C h ∆C F(top) F(bottom) m(C h) J 10-2Ri ∆t 109DV 109M

1.500 02 0.200 40 1.052 483 1.060 146 1.548 57 84.501 4.2167 7.8 1.4978 1.445

1.500 07 0.101 79 1.164 908 1.175 893 1.567 00 70.114 6.8880 7.6 0.5712 0.943

0.500 00 0.102 84 1.051 994 1.064 114 0.506 55 86.626 8.4235 8.9 0.7937 1.278

0.999 99 0.102 40 1.109 687 1.121 151 1.027 34 77.260 7.5449 6.8 0.6545 1.129

1.200 01 0.102 23 1.132 022 1.143 297 1.240 70 74.213 7.2594 9.1 0.6163 1.056

1.500 01 0.100 09 1.164 933 1.175 775 1.566 93 68.534 6.8472 10.8 0.5706 0.942

a Units of C h and ∆C are mol‚dm-3, of m(C h ) are mol‚kg-1, of 10-2 Ri are mol-1‚dm3, of F are g‚cm-3, of ∆t are s, and of 109DV and 109M are m2‚s-1. The density values were measured with a Mettler-Parr DMA/40 vibrating tube densimeter. Cell C-1235-H-11 was used for the diffusion measurements; for this cell the path length inside the cell is a ) 2.5057 cm and the magnification factor is 1.76070. b The stock solution was prepared using Baker “Analyzed” Na2SO4(s) and purified water and was filtered through a 0.2 µm Corning “Low Extractable” Membrane Filtering Unit before being used for making solutions for diffusion experiments. c The stock solution was prepared using recrystallized Baker “Analyzed” Na2SO4 that was separated from the mother liquor by centrifugal draining, and purified water, and was then filtered through a 0.2 µm Corning “Low Extractable” Membrane Filtering Unit before being used for making solutions for diffusion experiments.

All apparent masses were converted to masses by using buoyancy corrections. Ternary solutions with molarity fractions of z1 ) 0.900 00, 0.500 00, and 0.250 00 were prepared using dried OmniPur NaCl, whereas for the binary solution experiment and for the ternary solutions with z1 ) 0.749 93 we used dried Mallinckrodt Analytical Reagent NaCl. Two stock solutions of Na2SO4(aq) were prepared directly from anhydrous Baker “Analyzed” Na2SO4(s) and purified water. An additional stock solution was used for four of the binary solution measurements and was prepared similarly from Na2SO4‚10H2O(cr). That Na2SO4‚10H2O(cr) was obtained by recrystallization of Baker “Analyzed” Na2SO4(s) followed by centrifugal draining. All of these Na2SO4(aq) stock solutions were filtered through a prewashed 0.2 µm Corning “Low Extractable” Membrane Filtering Unit before use. The densities of all Na2SO4(aq) stock solutions were measured at (298.15 ( 0.005) K using a Mettler-Parr DMA/ 40 vibrating tube densimeter which was interfaced to an Apple computer for signal averaging and increased precision. The molar concentration of each Na2SO4(aq) stock solution was calculated from its measured density using eq 6 of Rard et al.11 That equation is valid up to 2.6292 mol‚dm-3, which is well into the supersaturated region. We estimate that the stock solution molarities and molalities derived from density measurements are accurate to 0.02% or better and that the consistency of concentrations for solutions prepared from different Na2SO4(aq) stock solutions is better than 0.01%. However, since all eight solutions for any particular ternary solution composition were prepared from the same stock solution, they are internally more consistent than this. The densities of all solutions used for diffusion experiments were also measured with this vibrating tube densimeter. At each given z1, the eight densities from the four solution pairs were represented by the linear Taylor series expansion,12,13

F ) Fj + H1(C1 - 〈C h 1〉) + H2(C2 - 〈C h 2〉)

(2)

using the method of least-squares. C1 and C2 are the concentrations of NaCl and Na2SO4, respectively, for each individual solution, 〈C h 1〉 and 〈C h 2〉 are the corresponding

h 2 for all four overall concentration averages of C h 1 and C solution pairs at the same overall composition, the Hi are least-squares parameters, and Fj is a least-squares parameter representing the density of a solution with molar concentrations corresponding to 〈C h 1〉 and 〈C h 2〉. These H1 and H2 parameters were used for calculating the partial molar volumes V h i of the two solutes and water,11,13 given in Table 3, which in turn were used to convert the experimentally based volume-fixed diffusion coefficients (Dij)V to solventfixed ones (Dij)0. The appropriate equation is

h 1〉 - H2〈C h 2〉) V h i ) (Mi - Hi)/(Fj - H1〈C

(3)

where Mi is the molar mass of component i and H0 ) 0 for the solvent. This equation is a generalization of the separate equations used previously11 to calculate the partial molar volumes of the solutes and solvent. Calculations for Ternary Solutions The complete description of diffusion of solutes in a ternary solution under isothermal and isobaric conditions requires four diffusion coefficients Dij, where i and j ) 1 or 2.6,35 The main-term diffusion coefficients Dii describe the flow of solute i due to its own concentration gradient, and the cross-term diffusion coefficients Dij (i * j) describe the flow of solute i due to a gradient of solute j. Under our experimental conditions of relatively small ∆Ci, these Dij are in the volume-fixed reference frame36 and are denoted as (Dij)V. Experimental values of the (Dij)V and Hi coefficients of eq 2 were used to test the static and dynamic stability of all our ternary solution diffusion boundaries.37-39 All of the diffusion boundaries were stable. Table 1 contains all concentration information for solutions used in our binary solution diffusion experiments, along with the densities and other experimental and derived information, and Table 2 contains the same type of information for the ternary solution experiments. Quantities reported (some are defined in the Experimental Section) are the C h i for both solutes; ∆t, the starting time correction which is added to the recorded “clock” times to correct them to the times corresponding to diffusion from an infinitely sharp boundary; and the reduced height-area ratio DA.24,40 The ternary solution diffusion coefficients

Journal of Chemical and Engineering Data, Vol. 45, No. 5, 2000 939 Table 2. Compositions and Results for Ternary Solution Diffusion Experiments for {z1NaCl + (1 - z1)Na2SO4}(aq) Solutions at 298.15 K and at 〈C h T〉 ) 1.500 mol‚dm-3 Measured with Rayleigh Interferometrya z1 ) 0.900 00

z1 ) 0.749 93

quantity

expt 1

expt 2

expt 3

expt 4

expt 1

expt 2

expt 3

expt 4

C h1 C h2 ∆C1 ∆C2 J(exptl) J(calcd) R1 ∆t 109DA(exptl) 109DA(calcd) F(top) F(bottom)

1.349 086 0.149 898 -0.000 110 0.098 435 79.3665 79.461 -0.000 57 9.9 0.8465 0.8468 1.062 313 1.073 676

1.349 124 0.149 905 0.195 434 0.000 011 80.767 80.863 0.999 89 11.2 1.4155 1.4147 1.064 326 1.071 699

1.349 164 0.149 909 0.039 155 0.078 770 79.916 79.822 0.202 94 9.1 0.9320 0.9305 1.062 706 1.073 310

1.349 058 0.149 894 0.156 426 0.019 706 80.728 80.632 0.802 60 12.8 1.2644 1.2646 1.063 924 1.072 092

1.124 472 0.374 968 0.000 146 0.102 602 80.529 80.565 0.000 73 8.2 0.7795 0.7786 1.079 600 1.091 316

1.124 475 0.374 973 0.199 908 -0.000 006 80.295 80.329 1.000 06 8.0 1.3123 1.3094 1.081 782 1.089 166

1.124 462 0.374 970 0.040 078 0.082 078 80.545 80.508 0.200 05 11.3 0.8553 0.85485 1.080 050 1.090 886

1.124 476 0.374 974 0.159 929 0.020 514 80.397 80.364 0.799 71 8.7 1.1653 1.1664 1.081 349 1.089 599

z1 ) 0.500 00

z1 ) 0.250 00

quantity

expt 1

expt 2

expt 3

expt 4

expt 1

expt 2

expt 3

expt 4

C h1 C h2 ∆C1 ∆C2 J(exptl) J(calcd) R1 ∆t 109DA(exptl) 109DA(calcd) F(top) F(bottom)

0.749 908 0.749 918 -0.000 080 0.107 914 80.775 80.751 -0.000 38 7.9 0.6894 0.6889 1.108 134 1.120 188

0.749 967 0.749 967 0.209 461 0.000 062 80.616 80.535 0.999 43 10.3 1.1665 1.1656 1.110 418 1.117 927

0.749 972 0.749 986 0.041 897 0.086 414 80.783 80.787 0.199 29 7.6 0.7583 0.7572 1.108 598 1.119 742

0.749 966 0.749 975 0.167 639 0.021 728 80.582 80.683 0.798 41 11.1 1.0370 1.0364 1.109 930 1.118 392

0.375 052 1.125 151 -0.000 030 0.112 585 80.753 80.718 -0.000 14 10.3 0.6210 0.6208 1.136 306 1.148 641

0.375 064 1.125 192 0.212 432 0.000 043 78.437 78.428 0.999 61 9.7 1.0630 1.0629 1.138 781 1.146 190

0.375 045 1.125 174 0.042 482 0.090 186 80.301 80.345 0.195 13 10.6 0.6824 0.6822 1.136 798 1.148 177

0.375 038 1.125 119 0.169 932 0.022 479 78.831 78.831 0.795 53 10.3 0.9400 0.9407 1.138 300 1.146 655

a Units of C h i and ∆Ci are mol‚dm-3, of ∆t are s, of 109DA are m2‚s-1, and of F are g‚cm-3. Densities were measured using a Mettler-Parr DMA/40 vibrating tube densimeter. Cell C-1235-H-11 was used; the path length inside this cell is a ) 2.5057 cm, and the magnification factor is 1.76070.

Table 3. Results from Ternary Solution Mutual Diffusion Coefficient and Density Measurements for {z1NaCl + (1 z1)Na2SO4}(aq) Solutions at 〈C h T〉 ) 1.500 mol‚dm-3 and 298.15 K Using Rayleigh Interferometrya quantity

z1 ) 0.900 00

z1 ) 0.749 93

z1 ) 0.500 00

z1 ) 0.250 00

〈C h T〉 〈C h 1〉 〈C h 2〉 〈C h 0〉 m1(〈C h 1〉,〈C h 2〉) m2(〈C h 1〉,〈C h 2〉) 10-2R1 10-2R2 Fj H1 H2 s(F) s(Fj) V h1 V h2 V h0 10-9σ+ 10-9σ10-2SA 109(D11)V 109(D12)V 109(D21)V 109(D22)V 109(D11)0 109(D12)0 109(D21)0 109(D22)0

1.499 0095 1.349 108 0.149 9015 53.7248 1.393 896 0.154 878 4.137 14 8.077 06 1.068 006 37.707 ( 0.065 115.625 ( 0.129 0.000 011 0.000 004 20.740 26.417 18.019 0.671 33 1.321 69 -77.74 1.5018 ( 0.0007 0.2970 ( 0.0011 -0.0305 ( 0.0003 0.7444 ( 0.0004 1.5441 0.3330 -0.0258 0.7484

1.499 442 1.124 471 0.374 971 53.6483 1.163 458 0.387 972 4.018 53 7.846 47 1.085 468 36.939 ( 0.049 114.081 ( 0.095 0.000 009 0.000 003 21.479 27.9235 17.9945 0.679 59 1.431 38 -82.08 1.4964 ( 0.0012 0.3039 ( 0.0017 -0.0655 ( 0.0005 0.6737 ( 0.0007 1.5317 0.3334 -0.0537 0.6835

1.499 915 0.749 953 0.749 9615 53.4998 0.778 110 0.778 118 3.842 65 7.485 78 1.114 166 35.880 ( 0.055 111.687 ( 0.107 0.000 012 0.000 004 22.485 30.244 17.9525 0.715 36 1.589 46 -88.12 1.4278 ( 0.0012 0.2603 ( 0.0017 -0.0917 ( 0.0005 0.5993 ( 0.0006 1.4507 0.2790 -0.0688 0.6180

1.500 209 0.375 050 1.125 159 53.3295 0.390 373 1.171 128 3.690 45 7.170 48 1.142 481 34.777 ( 0.053 109.633 ( 0.101 0.000 011 0.000 004 23.523 32.208 17.906 0.781 535 1.696 33 -94.65 1.2953 ( 0.0008 0.1372 ( 0.0011 -0.0809 ( 0.0003 0.5738 ( 0.0004 1.3062 0.1457 -0.0481 0.5994

a Units of 〈C h i〉 are mol‚dm-3, of mi(〈C h 1〉,〈C h 2〉) are mol‚kg-1, of 10-2Ri are mol-1‚dm3, of Fj, s(F), and s(Fj) are g‚cm-3, of Hi are g‚mol-1, of V h i are cm3‚mol-1, of 10-9σ+ and 10-9σ- are m-2‚s, of 10-2SA are m-1‚s1/2, and of 109(Dij)V and 109(Dij)0 are m2‚s-1. Here s(F) and s(Fj) are the standard deviations of the density fit and of Fj, respectively. The quantity z1 is the solute molarity fraction of NaCl, the total solute molarity is 〈C h T〉 ) 〈C h 1〉 + 〈C h 2〉, and 〈C h 0〉 is the molar concentration of water in the solution. To obtain densities from eq 2 in units of g‚cm-3 when Ci and 〈C h i〉 are in units of mol‚dm-3, divide the listed values of Hi by 103. The “(” value given immediately to the right of each (Dij)V value is its standard error as calculated from the data reduction algorithm using standard propagation of error methods.

(Dij)V were obtained using TFIT exactly as described in Section 3 of ref 11.

Experimental and calculated J values are both reported in Table 2. These J(calcd) were obtained from the ∆Ci and

940

Journal of Chemical and Engineering Data, Vol. 45, No. 5, 2000

the least-squares parameters Ri, using the second equality of eq 1. Calculation of the DA(exptl) values was performed as described on page 4193 of ref 11. The DA(calcd) for each experiment was obtained as described elsewhere11,24 using the R1 values of that experiment and the four least-squares Rayleigh parameters a, b, s1, and s2 appropriate to that overall composition, where s1 ) xσ+ and s2 ) xσ-. The quantities σ+ and σ- are defined in terms of the (Dij)V by eqs 12 and 13 of ref 11 and are the reciprocals of the eigenvalues of the diffusion coefficient matrix. The a and b parameters are defined in terms of the (Dij)V and Ri by eqs 8 and 9 of ref 24. Comparing these calculated DA values with the corresponding experimental ones provides a measure of the internal consistency of the four experiments at each overall composition. Agreement between DA(exptl) and DA(calcd) values is generally very good, |DA(exptl) - DA(calcd)| e 0.0015 × 10-9 m2‚s-1, except for the second experiment for the composition with z1 ) 0.749 93, where the difference is somewhat larger. Results Binary Solutions. Isothermal diffusion in a binary solution is characterized by a single, concentration dependent, volume-fixed diffusion coefficient DV. Table 1 contains the experimental results for the limiting binary solutions NaCl(aq) (z1 ) 1) and Na2SO4(aq) (z1 ) 0) at 298.15 K. Also given is the thermodynamic diffusion coefficient M ) DV/{d(mφ)/dm}, where φ is the molality-based or “practical” osmotic coefficient of the solution. These derivatives were evaluated at the molality m(C h ) corresponding to C h , using published equations for φ of Na2SO4(aq) and NaCl(aq).41,42 Newer and much more accurate extended Pitzer equations10 are available for the thermodynamic properties of both NaCl(aq) and Na2SO4(aq) as functions of temperature,43,44 which will allow more accurate values of the chemical potential derivatives {d(mφ)/dm} to be calculated at 298.15 K. However, we used the older empirical isothermal activity equations to facilitate comparison of our present values of M with previous results from our two laboratories.11,30,45,46 Rard and Miller45 reported DV for NaCl(aq) at 298.15 K from dilute solution to near saturation using Rayleigh interferometry and reviewed other published DV values. The original large-scale plot for their Figure 1 yields DV ) (1.4967 ( 0.002) × 10-9 m2‚s-1 at C h ) 1.5000 mol‚dm-3. Our experimental value of DV ) 1.4978 × 10-9 m2‚s-1 at C h ) 1.500 02 mol‚dm-3 agrees well within the uncertainty of the published values. h ) Our diffusion measurements for Na2SO4(aq) at C 1.500 07 mol‚dm-3, using Rayleigh interferometry, gave DV ) 0.5712 × 10-9 m2‚s-1. A value of DV ≈ (0.568 ( 0.001) × 10-9 m2‚s-1 at 1.500 mol‚dm-3 was estimated by graphical interpolation of a large-scale plot of the Rayleigh interferometric values of Rard and Miller,46 which were measured at Lawrence Livermore National Laboratory using a less precise diffusiometer. These two values disagree by 0.003 × 10-9 m2‚s-1 (0.5%), although no significant differences were observed at lower molarities.11,30 The present measurements were made using a stock solution of Na2SO4(aq) prepared from anhydrous Na2SO4(s) and purified water, whereas, in the earlier study,46 the Na2SO4 was recrystallized before use. Rard and Miller46 reported the pHs of their air-saturated solutions graphically, and their solution pHs were slightly alkaline (pH e 8.2) except at fairly low molalities where the pHs are determined mainly by the slight acidity of carbonic acid. Solutions of Na2SO4-

(aq) are expected to be very slightly alkaline due to a small amount of hydrolysis of the SO42- ion. In contrast, the stock solution of Na2SO4(aq) and the solutions prepared from it for the present diffusion study were found to be slightly acidic with pHs ∼ 6. We recrystallized a sample of our Baker “Analyzed” Na2SO4(s) and used it to prepare a new stock solution for some additional diffusion measurements for Na2SO4(aq) solutions. The solutions prepared from the recrystallized material had pHs nearly identical to those of solutions prepared from the unrecrystallized material. A diffusion experiment using samples prepared from the new stock solution, at C h ) 1.500 01 mol‚dm-3, gave DV ) 0.5706 × 10-9 m2‚s-1, which is lower than our first value of DV ) 0.5712 × 10-9 m2‚s-1 by 0.1%, but the difference between this value and that from the earlier experiment is not large. This agreement suggests that acidic impurities are probably not responsible for the moderate disagreement between our diffusion coefficients at C h ) 1.5000 mol‚dm-3 and those reported by Rard and Miller.46 We note that at pH ∼ 6 and pH ∼ 8 the molalities of hydrogen ion or hydroxide ion, respectively, are only about 10-6 mol‚dm-3, which are much too low for these ions to be altering the measured diffusion coefficients of Na2SO4(aq) at the concentration of interest. Some additional check experiments were made at three lower concentrations of Na2SO4(aq), where a comparison is also possible. Our new experimental measurement at C h ) 0.500 00 mol‚dm-3 gives DV ) 0.7937 × 10-9 m2‚s-1, Rard et al.11 obtained DV ) 0.7932 × 10-9 m2‚s-1 at C h ) 0.500 00 mol‚dm-3, Wendt47 obtained DV ) 0.7925 × 10-9 m2‚s-1 at C h ) 0.499 53 mol‚dm-3, and interpolation of the values of Rard and Miller46 to C h ) 0.5000 mol‚dm-3 gave11 DV ) (0.791 to 0.792) × 10-9 m2‚s-1. Similarly, our new measurement at C h ) 0.999 99 mol‚dm-3 gives DV ) 0.6545 × 10-9 2 -1 m ‚s , Albright et al.30 obtained DV ) 0.6546 × 10-9 m2‚s-1 at C h ) 1.000 02 mol‚dm-3, Wendt47 obtained DV ) 0.6539 × 10-9 m2‚s-1 at C h ) 0.99954 mol‚dm-3, and interpolation30 of the values of Rard and Miller46 to C h ) 1.0000 mol‚dm-3 gave DV ) (0.654 ( 0.001) × 10-9 m2‚s-1. All of these experimental values at both compositions are in very good agreement. At C h ) 1.200 01 mol‚dm-3 our new measurement gives DV ) 0.6163 × 10-9 m2‚s-1, which is slightly higher than the value of DV ∼ 0.614 × 10-9 m2‚s-1 estimated by interpolation of the results of Rard and Miller to C h ) 1.2000 mol‚dm-3. On the basis of previous experience with the Gosting diffusiometer, experimental DV are reproducible to about 0.03 to 0.05% for binary solutions, whereas the diffusiometer used by Rard and Miller46 was capable of a precision of 0.1-0.2%. There is thus a slight disagreement between the present results for Na2SO4(aq) and those of Rard and Miller at concentrations >1.0000 mol‚dm-3, which slightly exceeds the reported errors. In contrast, our two values measured with the Gosting diffusiometer agree very well at lower concentrations. Rard and Miller reported difficulties with crystallization of Na2SO4 while filling their cell with Na2SO4(aq) at their highest concentrations, since their laboratory was several degrees below 298 K when those experiments were performed. Consequently, they warmed their cell above room temperature before filling it, which could have given rise to slight systematic distortions in the baseline corrections. In our opinion, the combined measurements from this laboratory, those of Rard and Miller,46 and those of Wendt47 are the most accurate diffusion coefficients for Na2SO4(aq) at 298.15 K for concentrations up to 1.0000 mol‚dm-3 Na2SO4(aq), but at higher concentrations our new measurements reported in Table 1 are to

Journal of Chemical and Engineering Data, Vol. 45, No. 5, 2000 941 Table 4. Comparison of Calculated Errors of the Ternary Solution (Dij)V for {z1NaCl + (1 - z1)Na2SO4}(aq) Solutions at 〈C h T〉 ) 1.500 mol‚dm-3 and 298.15 Ka quantity

z1 ) 0.900 00

z1 ) 0.749 93

z1 ) 0.500 00

z1 ) 0.250 00

〈C h T〉 〈C h 1〉 〈C h 2〉 109δ(D11)Vb 109δ(D12)Vb 109δ(D21)Vb 109δ(D22)Vb 109δ(D11)Vc 109δ(D12)Vc 109δ(D21)Vc 109δ(D22)Vc

1.499 0095 1.349 108 0.149 9015 0.0007 0.0011 0.0003 0.0004 0.0033 0.0029 0.0010 0.0007

1.499 442 1.124 471 0.374 971 0.0012 0.0017 0.0005 0.0007 0.0008 0.0047 0.0011 0.0014

1.499 915 0.749 953 0.749 9615 0.0012 0.0017 0.0005 0.0006 0.0047 0.0125 0.0015 0.0041

1.500 209 0.375 050 1.125 159 0.0008 0.0011 0.0003 0.0004 0.0002 0.0058 0.0002 0.0017

a Units of 〈C h i〉 are mol‚dm-3 and of 109δ(Dij)V are m2‚s-1. The quantity z1 is the solute molarity fraction of NaCl in the mixed-electrolyte solutions. b The first set of errors was obtained with propagation of error equations using the variance-covariance matrix of the leastsquares parameters from the fits for all four experiments at each overall composition. c The second set of errors was obtained by the subset method. Reported uncertainties are n - 1 standard deviations.

be preferred. However, the presumed slight systematic errors in the four highest concentration diffusion coefficients of Rard and Miller46 are only 0.2-0.5% of the DV values. Ternary Solutions. Table 3 contains all the derived quantities for the four ternary solution compositions of the h T〉 ) 1.500 system NaCl + Na2SO4 + H2O at 298.15 K and 〈C mol‚dm-3. We report both the experimental volume-fixed (Dij)V and the derived solvent-fixed (Dij)0, which can be interconverted as described elsewhere.13,36 The reverse transformation, of the (Dij)0 to the (Dij)V, can be made using h 1〉,〈C h 2〉) and m2(〈C h 1〉,〈C h 2〉) eq 64 of Miller.6 The quantities m1(〈C are the molalities of NaCl and Na2SO4, respectively, corresponding to a solution having the molarities of both salts equal to the overall averages 〈C h 1〉 and 〈C h 2〉 of all four experiments at that overall composition. Another quantity reported in Table 3 is SA,48 which can be related to already defined quantities by

SA ) [D22 - D11 + (R1/R2)D12 - (R2/R1)D21]/[(D11D22 D12D21)(xσ+ + xσ-)] ) b(xσ+ - xσ-)

(4)

If |10-2SA| decreases below ∼(20 to 25) m-1‚s1/2, then the calculated standard errors of the Dij generally have significantly larger values than usual.24,49 Furthermore, if the σ+ and σ- values are nearly equal, the nonlinear leastsquares analysis of the diffusion data may not converge. Fortunately, σ+ and σ- differ by about a factor of 2 for our experiments and the |10-2SA| values range from (77.74 to 94.65) m-1‚s1/2, and no such computational difficulties were encountered. Reported uncertainties in the (Dij)V in Table 3 were obtained from the statistical analysis portion of TFIT using standard propagation of error methods. However, as discussed elsewhere, we believe the actual uncertainties are larger than these statistical uncertainties indicate.16,24,25,49,50 A more realistic “rule of thumb” estimate is that the actual errors are about four times larger than the statistical errors.20-24 Realistic errors for the (Dij)V may also be obtained using the data from various subsets of the diffusion experiments.50 These calculations were performed with the four possible three-experiment subsets of the R1 for each overall ternary solution composition. The results of this analysis are reported in Table 4, where the values of δ(Dij)V are “n - 1” standard deviations calculated from the four resulting subset values of each (Dij)V.

The calculated uncertainties of two of the (Dii)V for the z1 ) 0.500 00 case are somewhat larger than those given by the four times the statistical errors rule of thumb, but for the other three mixture compositions, there is an approximate agreement between the two different methods of estimating errors for the (Dij)V. This comparison suggests that the actual uncertainties of the two main-term (Dii)V coefficients are e0.005 × 10-9 m2‚s-1, of (D21)V are e0.002 × 10-9 m2‚s-1, and of (D12)V are generally e0.006 × 10-9 m2‚s-1. The exception is for (D12)V for the composition with z1 ) 0.500 00, where the uncertainty may be as large as 0.013 × 10-9 m2‚s-1. Discussion As z1 f 1, (D11)V f DV for NaCl(aq). Similarly, (D22)V f DV for Na2SO4(aq) as z1 f 0. Also, as z1 f 1, (D21)V f 0 because no Na2SO4(aq) is present to be transported by coupled diffusion, and (D12)V f 0 as z1 f 0 because no NaCl(aq) is present to be transported by coupled diffusion. Extrapolation of some of the (Dij)V to z1 ) 1 or z1 ) 0 yields significant new information. As z1 f 1, (D22)V f D*(SO42-), and as z1 f 0, (D11)V f D*(Cl-), where D*(SO42-) is the trace diffusion coefficient of the SO42- ion in a solution of 1.500 mol‚dm-3 NaCl(aq) and where D*(Cl-) is the trace diffusion coefficient of the Cl- ion in a solution with molarity 1.500 mol‚dm-3 Na2SO4(aq). Unlike the other extrapolated values, (D12)V as z1 f 1 and (D21)V as z1 f 0 are simply limiting values. The solid curves of Figures 1 and 2 show the trends in the main-term and cross-term (Dij)V, respectively, as functions of z1 at 〈C h T〉 ) (0, 0.500, 1.000, and 1.500) mol‚dm-3. Additional plots (not presented here), similar to Figures 1 and 2, were made at 1.500 mol‚dm-3 with the molarity composition fraction replaced by the ionic strength fraction, the equivalent fraction, and the ratio of the ionic molarity (osmolarity) of NaCl to the total ionic molarity. Extrapolated values of the cross-term coefficients, and of D*(Cl-) and of D*(SO42-), were obtained graphically from these four plots by two of us independently. Additional extrapolations were also used, as described below. The average of these and the DV values for the limiting binary solutions at molarities of 1.500 mol‚dm-3 are summarized in Table 5 and were used to extend the curves in Figures 1 and 2 to z1 ) 0 and 1. The experimental values of (D11)V and (D22)V are very smooth and regular functions of z1 and connect smoothly with the DV for their limiting binary solutions NaCl(aq) and Na2SO4(aq), respectively. Our extrapolations of (D22)V

942

Journal of Chemical and Engineering Data, Vol. 45, No. 5, 2000

Figure 1. Values of the volume-fixed mutual diffusion main-term coefficients (Dii)V at 298.15 K for NaCl + Na2SO4 + H2O at total concentration 〈C h T〉 ) (0.500, 1.000, and 1.500) mol‚dm-3, along with the corresponding values at infinite dilution (〈C h T〉 ) 0) from the Nernst-Hartley equation, as a function of the solute molarity fraction z1 of NaCl. Also plotted are values of DV for the limiting binary solutions NaCl(aq) (at z1 ) 1) and Na2SO4(aq) (at z1 ) 0) and extrapolated values of the trace diffusion coefficients D*(Cl-) and D*(SO42-). Symbols: O, 0, ], 4, (D11)V at (0, 0.500, 1.000, and 1.500) mol‚dm-3, respectively; b, 9, [, 2, (D22)V at (0, 0.500, 1.000, and 1.500) mol‚dm-3, respectively.

Figure 2. Values of the volume-fixed mutual diffusion cross-term coefficients (Dij)V at 298.15 K for NaCl + Na2SO4 + H2O at total concentration 〈C h T〉 ) (0.500, 1.000, and 1.500) mol‚dm-3, along with the corresponding values at infinite dilution (〈C h T〉 ) 0) from the Nernst-Hartley equation, as a function of the solute molarity fraction z1 of NaCl. Symbols: O, 0, ], 4, (D12)V at (0, 0.500, 1.000, and 1.500) mol‚dm-3, respectively; b, 9, [, 2, (D21)V at (0, 0.500, 1.000, and 1.500) mol‚dm-3, respectively. We revise our earlier estimate30 of (D12)V as z1 f 1 at concentration 〈C h T〉 ) 1.000 mol‚dm-3 from 0.204 × 10-9 m2‚s-1 to 0.215 × 10-9 m2‚s-1, on the basis of an examination of the variations of (D12)V with both 〈C h T〉 and z1.

to obtain D*(SO42-) should be fairly accurate. However, extrapolation of (D11)V to obtain D*(Cl-) has larger uncertainties because of the steep decrease of (D11)V as z1 f 0. We found that the difference (D11)V - (D12)V showed much less variation with the composition fraction of NaCl, and this function was therefore used to obtain more precise extrapolated values of D*(Cl-). Also, (D12)V goes through a maximum at z1 ∼ 0.8 and (D21)V has a minimum at z1 ∼ 0.4. Consequently, the direct extrapolation of (D12)V as z1 f 1 and of (D21)V as z1 f 0 would have greater uncertainties than our extrapolated values D*(SO42-), especially for (D21)V where the trend is not completely defined at low values of z1. The Nernst-Hartley equations for ternary electrolyte solutions6 predict that both D12/e1 and D21/e2 should be linear functions of e1, where e1 ) C1/(C1 + 2C2) and e2 ) 1 - e1 ) 2C2/(C1 + 2C2) are the equivalent fractions of the corresponding solutes. Although this linear relationship does not hold at our concentrations of 〈C h T〉 ) 1.500 mol‚dm-3, these two functions do not have the maxima or minima observed in the direct plots of (D12)V and (D21)V against z1. Thus, the extrapolation of (D21)V as z1 f 0 can be made more accurately using (D21)V/e2, and our extrapolated value in Table 5 was based on this plot and related plots of (D21)V/e2 against other composition fractions. However, the corresponding plots of (D12)V/e1 were less useful for obtaining the extrapolated value as z1 f 1. The (D21)V values generally become more negative as z1 f 0, and (D12)V usually becomes more positive as z1 f 1. Thus, coupled diffusion enhances the rate of diffusion of NaCl but reduces that of Na2SO4 in these solutions, and the magnitude of this coupling increases as the concentrations of the solutions are increased.

Our extrapolation technique, which uses mutual diffusion coefficients to determine trace diffusion coefficients, has the advantage over conventional methods that the solutions do not contain radioactive tracers or different isotopes. We are not aware of any direct determinations of D*(Cl-) or D*(SO42-) by conventional methods for the same ionic media used here. However, some comparisons are possible. A value of D*(SO42-) ) 0.816 × 10-9 m2‚s-1 was recommended by Mills and Lobo51 for solutions of Na2SO4(aq) at an ionic strength of I ) 1.476 mol‚dm-3. This value was taken from the then unpublished study of Weinga¨rtner et al.52 Agreement with our extrapolated value D*(SO42-) ) (0.805 ( 0.015) × 10-9 m2‚s-1 is excellent despite the different electrolyte media. Similarly, D*(Cl-) for several electrolyte solutions were tabulated by Mills and Lobo.51 Values of D*(Cl-) at I ) 4.5 mol‚dm-3 are needed, since this is the total ionic strength of our solutions at z1 ) 0. Our extrapolated value is D*(Cl-) ) (1.145 ( 0.02) × 10-9 m2‚s-1. The interpolated value of D*(Cl-) in NaCl(aq) from Mills and Lobo at I ) 4.5 mol‚dm-3, (1.16 ( 0.01) × 10-9 m2‚s-1, is essentially identical to our value in Na2SO4(aq). The values of D*(Cl-) are found to be virtually identical in solutions of NaCl(aq) and of Na2SO4(aq) at I ) 4.500 mol‚dm-3, which also happens at I ) 1.500 mol‚dm-3 and at I ) 3.000 mol‚dm-3.11,30 This good agreement also holds for D*(SO42-) in solutions of NaCl(aq) and of Na2SO4(aq) at both I ) 1.500 mol‚dm-3 (present study) and I ) 0.500 mol‚dm-3 and I ) 1.000 mol‚dm-3.11,30 Such close agreement between D*(Cl-) obtained in NaCl(aq) and Na2SO4(aq) solutions at the same stoichiometric ionic strengths, which occurs at all three ionic strengths from I ) (1.500 to

Journal of Chemical and Engineering Data, Vol. 45, No. 5, 2000 943 Table 5. Values of (Dij)V for {z1NaCl + (1 - z1)Na2SO4}(aq) Solutions as z1 f 0 and z1 f 1 at 〈C h T〉 ) 1.500 mol‚dm-3 and 298.15 Ka

a

quantity

109(Dij)V

interpretation

(D11)V as z1 f 1 (D12)V as z1 f 1 (D21)V as z1 f 1 (D22)V as z1 f 1 (D11)V as z1 f 0 (D12)V as z1 f 0 (D21)V as z1 f 0 (D22)V as z1 f 0

1.4978 ( 0.001 0.28 ( 0.02 0 0.805 ( 0.015 1.145 ( 0.02 0 -0.03 ( 0.02 0.5709b

DV(NaCl) at C1 ) 1.500 mol‚dm-3 extrapolated value at I ) 1.500 mol‚dm-3 by definition D*(SO42-) at I ) 1.500 mol‚dm-3 (extrapolated) D*(Cl-) at I ) 4.500 mol‚dm-3 (extrapolated) by definition extrapolated value at I ) 4.500 mol‚dm-3 DV(Na2SO4) at C2 ) 1.500 mol‚dm-3

Units of 109(Dij)V are m2‚s-1. b Average of our two experimental values given in Table 1.

4.500) mol‚dm-3, seems to imply that the ionic strengths of the Na2SO4(aq) solutions are not being reduced significantly by the formation of sodium sulfate ion pairs. This conclusion conflicts with some thermodynamic analyses53 which claim that fairly extensive ion pair formation occurs in these solutions. This good agreement may be more than coincidental and could imply that the trace diffusion coefficient of an anion Xn- in a solution of electrolyte MaYb(aq) can be reliably estimated from its value in the common cation solution of Ma′Xb′(aq) at the same ionic strength. However, mutual and isotope diffusion coefficient measurements for other ternary aqueous electrolyte systems are needed to test the generality of this observation. We note that the (D12)V and (D21)V values are of opposite sign and the difference between their values is approximately constant, (D12)V - (D21)V ) (0.317 ( 0.068) × 10-9 m2‚s-1, at 〈C h T〉 ) 1.500 mol‚dm-3 for the four values of z1. However, the differences are significantly larger for the two intermediate mixtures (with z1 ) 0.749 93 and 0.500 00) than for the other two mixtures which are closer to the limiting binary compositions (with z1 ) 0.900 00 and 0.250 00). This difference at 1.500 mol‚dm-3 is also somewhat larger than the difference found at 0.500 mol‚dm-3, (0.246 ( 0.018) × 10-9 m2‚s-1,11 or at 1.000 mol‚dm-3, (0.282 ( 0.034) × 10-9 m2‚s-1.30 Cross-term diffusion coefficients calculated from the Nernst-Hartley equation11 yield a comparable difference of (D12)V - (D21)V ) (0.266 ( 0.032) × 10-9 m2‚s-1 at infinite dilution. For 〈C h T〉 ) (0.500, 1.000, and 1.500) mol‚dm-3, (D12)V - (D21)V has a small maximum at intermediate values of z1, in contrast to the monotonic increase with z1 at infinite dilution that is predicted by the Nernst-Hartley equation. Figures 1 and 2 contain plots of the main-term (i ) j) and the cross-term (Dij)V (i * j), respectively, as functions of z1 at 〈C h T〉 ) (0, 0.500, 1.000, and 1.500) mol‚dm-3 and 298.15 K. The (Dij)V at 〈C h T〉 ) 0 (infinite dilution) were calculated using the ternary solution analogue of the Nernst-Hartley equation.6 The observed opposite signs for (D12)V and (D21)V at (0.500, 1.000, and 1.500) mol‚dm-3, see Figure 2, are predicted qualitatively by the Nernst-Hartley equation, which is Coulombically based. However, there are some sizable quantitative differences between the experimental (Dij)V and the Nernst-Hartley values, especially for the main-term (Dii)V, as can be seen in Figure 1. At 〈C h T〉 ) 1.500 mol‚dm-3, the Nernst-Hartley D11 are significantly higher than the experimental (D11)V by (0.186 to 0.678) × 10-9 m2‚s-1, and the Nernst-Hartley D22 are significantly higher than the experimental (D22)V by (0.351 to 0.633) × 10-9 m2‚s-1. In addition, there are significant qualitative differences. At infinite dilution, both D11 and D22 decrease monotonically with increasing z1. In contrast, (D22)V increases monotonically with increasing z1 at all three experimental concentrations of 〈C h T〉 ) (0.500

h T〉 ) to 1.500) mol‚dm-3. Although (D11)V values at both 〈C (0 and 0.500) mol‚dm-3 have similar qualitative trends with z1, at 〈C h T〉 ) 1.000 mol‚dm-3 its values increase from z1 ) 0 to ∼ 0.7 and then decrease at higher z1, and at 〈C h T〉 ) 1.500 mol‚dm-3 this maximum shifts to z1 ∼ 0.8. Also in contrast, at 〈C h T〉 ) 1.500 mol‚dm-3 the NernstHartley cross-term Dij (i * j) are closer to the experimental (Dij)V but are always smaller rather than larger. For D12, the Nernst-Hartley values are smaller than the experimental ones by (0.033 to 0.154) × 10-9 m2‚s-1. For D21, the Nernst-Hartley values are smaller (more negative) than the experimental (D21)V by (0.010 to 0.104) × 10-9 m2‚s-1. However, at 〈C h T〉 ) (0.500 and 1.000) mol‚dm-3, some of the experimental cross-term (D12)V are larger than the Nernst-Hartley values, whereas others are smaller. It is obvious that simple empirical corrections, such as dividing the Nernst-Hartley Dij values by the ratio of the viscosity of the solution to that of the solvent, will not bring their values into conformity with the experimental (Dij)V or the (Dij)0. Similarly, dividing the Nernst-Hartley Dij by the appropriate chemical potential derivatives generally brings the corrected Nernst-Hartley values into better agreement with the experimental values at lower concentrations. However, at high concentrations the resulting predicted Dij values may be considerably different than the experimental values even for relatively simple systems such as NaCl + SrCl2 + H2O at 298.15 K.17 There is no rigorous theoretical relationship between the Dij coefficients and the ratio of the viscosity of the solutions to that of the solvent, even for a binary solution diffusion coefficient DV, and thus this viscosity “correction” must be considered to be a purely empirical term.5,54 However, as discussed by Robinson and Stokes,55 including an adjustable hydration number into the model (that formally represents the number of waters bound to the ions), along with the viscosity and chemical potential derivative terms, yielded a much-improved representation of the concentration dependence of DV for several soluble strong electrolytes. The success of the model using all of the above factors suggests that the adjustable hydration number is compensating to a certain extent for some of the deficiencies of the model, including the approximate nature of the viscosity correction. Leaist and Al-Dhaher56 have generalized this model to common-ion mixed electrolyte solutions and applied it to analyze published diffusion coefficients for the NaCl + SrCl2 + H2O and NaCl + MgCl2 + H2O systems at 298.15 K.16,17,20-24 They were able to represent all of the qualitative features of the dependences of the (Dij)V upon 〈C h T〉 and zi. In certain composition regions the agreement between their model and experiment is quite good up to 〈C h T〉 ) 1.0 mol‚dm-3 or even to higher concentrations, which implies that their model will be useful for predicting diffusion coefficients of mixed electrolyte solutions. However, this

944

Journal of Chemical and Engineering Data, Vol. 45, No. 5, 2000

approach has not yet been applied to any common ion sodium salt mixtures such as NaCl + Na2SO4 + H2O. Three of the diffusion coefficients for NaCl + Na2SO4 + H2O, (D11)V, (D22)V, and (D21)V, form families of curves that vary slowly and systematically with increasing 〈C h T〉. Thus, it should be possible to accurately estimate values of their diffusion coefficients at intermediate concentrations by interpolation. In contrast, the Nernst-Hartley infinite dilution values of D12 have a significantly different qualitative behavior than the experimental (D12)V. The NernstHartley curve does not predict the observed maximum and crosses all of the experimental (D12)V curves. Therefore, interpolation to intermediate concentrations could yield more uncertain results for (D12)V in certain regions of z1 for 〈C h T〉 < 1.000 mol‚dm-3. There are some significant changes with concentration in the variation of the (D12)V with z1 at constant 〈C h T〉. At infinite dilution the Nernst-Hartley values of both (D12)V and (D21)V increase monotonically with increasing values of z1. However, at 〈C h T〉 ) 0.500 mol‚dm-3 the values of (D12)V approach a maximum at z1 ∼ 0.5, at 〈C h T〉 ) 1.000 mol‚dm-3 the maximum is at z1 ∼ 0.7, and this maximum shifts to z1 ∼ 0.8 at 〈C h T〉 ) 1.500 mol‚dm-3. Neither this maximum nor its shift with increasing concentration is predicted by the Nernst-Hartley equations. At infinite dilution the values of (D21)V also increase monotonically as z1 increases. However, at finite concentrations the values of (D21)V at any fixed low z1 begin to increase with increasing 〈C h T〉, but by 〈C h T〉 ) 1.000 mol‚dm-3 values of (D21)V are essentially constant from z1 ) 0 to ∼ 0.25. In contrast, at 〈C h T〉 ) 1.500 mol‚dm-3 the values of (D21)V go through a minimum at z1 ∼ 0.4 and then increase at larger values of z1. It is not possible to accurately assess how rapidly (D21)V increases at lower values of z1, since there is only a single composition point to characterize this region, but the presence of the upturn suggests that values of (D21)V at low z1 will switch from negative to positive somewhere between 〈C h T〉 ) 2 and 3 mol‚dm-3. However, because of solubility limitations for the precipitation of Na2SO4‚10H2O(cr), it will probably not be possible to investigate this composition region using presently available experimental techniques. Acknowledgment We thank Joanne L. Levatin (Lawrence Livermore National Laboratory) for adapting some of our older computer programs to meet our current requirements. Literature Cited (1) Anderson, D. E.; Graf, D. L. Ionic Diffusion in Naturally-Occurring Aqueous Solutions: Use of Activity Coefficients in TransitionState Models. Geochim. Cosmochim. Acta 1978, 42, 251-262. (2) Felmy, A. R.; Weare, J. H. Calculation of Multicomponent Ionic Diffusion from Zero to High Concentration: I. The System NaK-Ca-Mg-Cl-SO4-H2O at 25 °C. Geochim. Cosmochim. Acta 1991, 55, 113-131. (3) Noulty, R. A.; Leaist, D. G. Diffusion in Aqueous Copper Sulfate and Copper Sulfate-Sulfuric Acid Solutions. J. Solution Chem. 1987, 16, 813-825. (4) Steefel, C. I.; Lichtner, P. C. Diffusion and Reaction in Rock Matrix Bordering a Hyperalkaline Fluid-Filled Fracture. Geochim. Cosmochim. Acta 1994, 58, 3595-3612. (5) Miller, D. G. Application of Irreversible Thermodynamics to Electrolyte Solutions. I. Determination of Ionic Transport Coefficients lij for Isothermal Vector Transport Processes in Binary Electrolyte Systems. J. Phys. Chem. 1966, 70, 2639-2659. (6) Miller, D. G. Application of Irreversible Thermodynamics to Electrolyte Solutions. II. Ionic Coefficients lij for Isothermal Vector Transport Processes in Ternary Systems. J. Phys. Chem. 1967, 71, 616-632. (7) Miller, D. G. Application of Irreversible Thermodynamics to Electrolyte Solutions. III. Equations for Isothermal Vector Trans-

(8)

(9) (10) (11)

(12) (13)

(14)

(15) (16) (17) (18)

(19) (20) (21)

(22)

(23) (24)

(25)

(26)

(27) (28) (29) (30)

(31)

port Processes in n-Component Systems. J. Phys. Chem. 1967, 71, 3588-3592. Miller, D. G. Explicit Relations of Velocity Correlation Coefficients to Onsager lij's, to Experimental Quantities, and to Infinite Dilution Limiting Laws for Binary Electrolyte Solutions. J. Phys. Chem. 1981, 85, 1137-1146. The terms r12c/(103F2) and r22c/(103F2) should be in front of the left-hand square brackets of eqs A1 and A2, respectively, and the term in front of the square bracket of eq A3 should be c02/(103F2). Zhong, E. C.; Friedman, H. L. Self-Diffusion and Distinct Diffusion of Ions in Solution. J. Phys. Chem. 1988, 92, 1685-1692. Pitzer, K. S. Ion Interaction Approach: Theory and Data Correlation. In Activity Coefficients in Electrolyte Solutions, 2nd ed.; Pitzer, K. S., Ed.; CRC Press: Boca Raton, FL, 1991; Chapter 3. Rard, J. A.; Albright, J. G.; Miller, D. G.; Zeidler, M. E. Ternary Mutual Diffusion Coefficients and Densities of the System {z1NaCl + (1 - z1)Na2SO4}(aq) at 298.15 K and a Total Molarity of 0.5000 mol dm-3. J. Chem. Soc., Faraday Trans. 1996, 92, 41874197. Dunlop, P. J. Data for Diffusion in Concentrated Solution of the System NaCl-KCl-H2O at 25°. A Test of the Onsager Reciprocal Relation for this Composition. J. Phys. Chem. 1959, 63, 612-615. Dunlop, P. J.; Gosting, L. J. Use of Diffusion and Thermodynamic Data to Test the Onsager Reciprocal Relation for Isothermal Diffusion in the System NaCl-KCl-H2O at 25°. J. Phys. Chem. 1959, 63, 86-93. O’Donnell, I. J.; Gosting, L. J. The Concentration Dependence of the Four Diffusion Coefficients of the System NaCl-KCl-H2O at 25 °C. In The Structure of Electrolytic Solutions; Hamer, W. J., Ed.; John Wiley: New York, 1959; pp 160-182. Kim, H. Diffusion Coefficients for One Composition of the System H2O-NaCl-KCl. J. Chem. Eng. Data 1982, 27, 255-256. Rard, J. A.; Miller, D. G. Ternary Mutual Diffusion Coefficients of NaCl-SrCl2-H2O at 25 °C. 1. Total Concentrations of 0.5 and 1.0 mol‚dm-3. J. Phys. Chem. 1987, 91, 4614-4620. Rard, J. A.; Miller, D. G. Ternary Mutual Diffusion Coefficients of NaCl-SrCl2-H2O at 25 °C. 2. Total Concentrations of 2.0 and 3.0 mol‚dm-3. J. Phys. Chem. 1988, 92, 6133-6140. Miller, D. G.; Ting, A. W.; Rard, J. A.; Eppstein, L. B. Ternary Diffusion Coefficients of the Brine Systems NaCl (0.5 M)-Na2SO4 (0.5 M)-H2O and NaCl (0.489 M)-MgCl2 (0.051 M)-H2O (Seawater Composition) at 25 °C. Geochim. Cosmochim. Acta 1986, 50, 2397-2403. Leaist, D. G. Ternary Diffusion in Aqueous NaCl + MgCl2 Solutions at 25 °C. Electrochim. Acta 1988, 33, 795-799. Albright, J. G.; Mathew, R.; Miller, D. G.; Rard, J. A. Isothermal Diffusion Coefficients for NaCl-MgCl2-H2O at 25 °C. 1. Solute Concentration Ratio of 3:1. J. Phys. Chem. 1989, 93, 2176-2180. Paduano, L.; Mathew, R.; Albright, J. G.; Miller, D. G.; Rard, J. A. Isothermal Diffusion Coefficients for NaCl-MgCl2-H2O at 25 °C. 2. Low Concentrations of NaCl with a Wide Range of MgCl2 Concentrations. J. Phys. Chem. 1989, 93, 4366-4370. Mathew, R.; Paduano, L.; Albright, J. G.; Miller, D. G.; Rard, J. A. Isothermal Diffusion Coefficients for NaCl-MgCl2-H2O at 25 °C. 3. Low MgCl2 Concentrations with a Wide Range of NaCl Concentrations. J. Phys. Chem. 1989, 93, 4370-4374. Mathew, R.; Albright, J. G.; Miller, D. G.; Rard, J. A. Isothermal Diffusion Coefficients for NaCl-MgCl2-H2O at 25 °C. 4. Solute Concentration Ratio of 1:3. J. Phys. Chem. 1990, 94, 6875-6878. Miller, D. G.; Albright, J. G.; Mathew, R.; Lee, C. M.; Rard, J. A.; Eppstein, L. B. Isothermal Diffusion Coefficients of NaCl-MgCl2H2O at 25 °C. 5. Solute Concentration Ratio of 1:1 and Some Rayleigh Results. J. Phys. Chem. 1993, 97, 3885-3899. Miller, D. G.; Sartorio, R.; Paduano, L.; Rard, J. A.; Albright, J. G. Effects of Different Sized Concentration Differences Across Free Diffusion Boundaries and Comparison of Gouy and Rayleigh Diffusion Measurements Using NaCl-KCl-H2O. J. Solution Chem. 1996, 25, 1185-1211. Albright, J. G.; Mathew, R.; Miller, D. G. Measurement of Binary and Ternary Mutual Diffusion Coefficients of Aqueous Sodium and Potassium Bicarbonate Solutions at 25 °C. J. Phys. Chem. 1987, 91, 210-215. Hao, L.; Leaist, D. G. Interdiffusion Without a Common Ion in Aqueous NaCl-MgSO4 and LiCl-NaOH Mixed Electrolytes. J. Solution Chem. 1995, 24, 523-535. Leaist, D. G. Moments Analysis of Restricted Ternary Diffusion: Sodium Sulfite + Sodium Hydroxide + Water. Can. J. Chem. 1985, 63, 2933-2939. Deng, Z.; Leaist, D. G. Ternary Mutual Diffusion Coefficients of MgCl2 + MgSO4 + H2O and Na2SO4 + MgSO4 + H2O from Taylor Dispersion Profiles. Can. J. Chem. 1991, 69, 1548-1553. Albright, J. G.; Gillespie, S. M.; Rard, J. A.; Miller, D. G. Ternary Solution Mutual Diffusion Coefficients and Densities of Aqueous Mixtures of NaCl and Na2SO4 at 298.15 K for Six Different Solute Fractions at a Total Molarity of 1.000 mol‚dm-3. J. Chem. Eng. Data 1998, 43, 668-675. Gosting, L. J.; Kim, H.; Loewenstein, M. A.; Reinfelds, G.; Revzin, A. A Versatile Optical Diffusiometer Including a Large Optical Bench of New Design. Rev. Sci. Instrum. 1973, 44, 1602-1609.

Journal of Chemical and Engineering Data, Vol. 45, No. 5, 2000 945 (32) Miller, D. G.; Albright, J. G. Optical Methods. In Measurement of the Transport Properties of Fluids; Wakeham, W. A., Nagashima, A., Sengers, J. V., Eds.; Experimental Thermodynamics, Vol. III; Blackwell Scientific Publications: Oxford, 1991; pp 272294 (references on pp 316-320). (33) Miller, D. G. A Method for Obtaining Multicomponent Diffusion Coefficients Directly from Rayleigh and Gouy Fringe Position Data. J. Phys. Chem. 1988, 92, 4222-4226. (34) Rard, J. A. Isopiestic Determination of the Osmotic Coefficients of Lu2(SO4)3(aq) and H2SO4(aq) at the Temperature T ) 298.15 K, and Review and Revision of the Thermodynamic Properties of Lu2(SO4)3(aq) and Lu2(SO4)3‚8H2O(cr). J. Chem. Thermodyn. 1996, 28, 83-110. (35) Onsager, L. Theories and Problems of Liquid Diffusion. Ann. N. Y. Acad. Sci. 1945, 46, 241-265. (36) Woolf, L. A.; Miller, D. G.; Gosting, L. J. Isothermal Diffusion Measurements on the System H2O-Glycine-KCl at 25°; Tests of the Onsager Reciprocal Relation. J. Am. Chem. Soc. 1962, 84, 317-331. (37) Vitagliano, P. L.; Della Volpe, C.; Vitagliano, V. Gravitational Instabilities in Free Diffusion Boundaries. J. Solution Chem. 1984, 13, 549-562. (38) Miller, D. G.; Vitagliano, V. Experimental Test of McDougall’s Theory for the Onset of Convective Instabilities in Isothermal Ternary Systems. J. Phys. Chem. 1986, 90, 1706-1717. (39) Vitagliano, V.; Borriello, G.; Della Volpe, C.; Ortona, O. Instabilities in Free Diffusion Boundaries of NaCl-Sucrose-H2O Solutions at 25 °C. J. Solution Chem. 1986, 15, 811-826. (40) Dunlop, P. J.; Gosting, L. J. Interacting Flows in Liquid Diffusion: Expressions for the Solute Concentration Curves in Free Diffusion, and their Use in Interpreting Gouy Diffusioneter Data for Aqueous Three-Component Systems. J. Am. Chem. Soc. 1955, 77, 5238-5249. (41) Rard, J. A.; Miller, D. G. Isopiestic Determination of the Osmotic Coefficients of Aqueous Na2SO4, MgSO4, and Na2SO4-MgSO4 at 25 °C. J. Chem. Eng. Data 1981, 26, 33-38. (42) Hamer, W. J.; Wu, Y.-C. Osmotic Coefficients and Mean Activity Coefficients of Uni-Univalent Electrolytes in Water at 25 °C. J. Phys. Chem. Ref. Data 1972, 1, 1047-1099. (43) Archer, D. G. Thermodynamic Properties of the NaCl + H2O System. II. Thermodynamic Properties of NaCl(aq), NaCl‚2H2O(cr), and Phase Equilibria. J. Phys. Chem. Ref. Data 1992, 21, 793-829. (44) Rard, J. A.; Clegg, S. L.; Palmer, D. A. Isopiestic Determination of the Osmotic Coefficients of Na2SO4(aq) at 25 and 50 °C, and Representation with Ion-Interaction (Pitzer) and Mole Fraction Thermodynamic Models. J. Solution Chem. 2000, 29, 1-49. (45) Rard, J. A.; Miller, D. G. The Mutual Diffusion Coefficients of NaCl-H2O and CaCl2-H2O at 25 °C from Rayleigh Interferometry. J. Solution Chem. 1979, 8, 701-716.

(46) Rard, J. A.; Miller, D. G. The Mutual Diffusion Coefficients of Na2SO4-H2O and MgSO4-H2O at 25 °C from Rayleigh Interferometry. J. Solution Chem. 1979, 8, 755-766. (47) Wendt, R. P. Studies of Isothermal Diffusion at 25° in the System Water-Sodium Sulfate-Sulfuric Acid and Tests of the Onsager Relation. J. Phys. Chem. 1962, 66, 1279-1288. (48) Fujita, H.; Gosting, L. J. An Exact Solution of the Equations for Free Diffusion in Three-component Systems with Interacting Flows, and its Use in Evaluation of the Diffusion Coefficients. J. Am. Chem. Soc. 1956, 78 (8), 1099-1106. (49) Rard, J. A.; Miller, D. G. Ternary Mutual Diffusion Coefficients of ZnCl2-KCl-H2O at 25 °C by Rayleigh Interferometry. J. Solution Chem. 1990, 19, 129-148. (50) Miller, D. G.; Paduano, L.; Sartorio, R.; Albright, J. G. Analysis of Gouy Fringe Data and Comparison of Rayleigh and Gouy Optical Diffusion Measurements Using the System Raffinose (0.015 M)-KCl (0.5 M)-H2O at 25 °C. J. Phys. Chem. 1994, 98, 13745-13754. (51) Mills, R.; Lobo, V. M. M. Self-diffusion in Electrolyte Solutions; Elsevier: Amsterdam, 1989. (52) Weinga¨rtner, H.; Price, W. E.; Edge, A. V. J.; Mills, R. Transport Measurements in Aqueous Sodium Sulfate. Evidence for LikeIon Pairs in Concentrated Solutions. J. Phys. Chem. 1993, 97, 6289-6291. (53) Johnson, K. S.; Pytkowicz, R. M. Ion Association and Activity Coefficients in Multicomponent Solutions. In Activity Coefficients in Electrolyte Solutions; Pytkowicz, R. M., Ed.; CRC Press: Boca Raton, FL, 1979; Vol. II, Chapter 1. (54) Miller, D. G. Certain Transport Properties of Binary Electrolyte Solutions and Their Relation to the Thermodynamics of Irreversible Processes. J. Phys. Chem. 1960, 64, 1598-1599. (55) Robinson, R. A.; Stokes, R. H. Electrolyte Solutions, 2nd ed. (Revised); Butterworth: London, 1965. (56) Leaist, D. G.; Al-Dhaher, F. F. Predicting the Diffusion Coefficients of Concentrated Mixed Electrolyte Solutions from Binary Solution Data. NaCl + MgCl2 + H2O and NaCl + SrCl2 + H2O at 25 °C. J. Chem. Eng. Data 2000, 45, 308-314.

Received for review May 1, 2000. Accepted June 26, 2000. Support from the Chemistry Department of Texas Christian University and from NASA Biotechnology Program Grant NAG8-1356 for O.A. and J.G.A. is gratefully acknowledged. The work of J.A.R. and D.G.M. was performed under the auspices of the Office of Basic Energy Sciences, Division of Geosciences, of the U.S. Department of Energy by University of California Lawrence Livermore National Laboratory under Contract No. W-7405-Eng-48.

JE000134S